The Classification of Wine Based on PCA and ANN

摘要

The qualitative identification of different wine through electronic nose is introduced. Principal component analysis (PCA) and artificial neural network (ANN) are adopted to realize the identification. An improved Back Propagation neural network (BP) algorithm, nearest neighbor-clustering Radial Basis Function (RBF) algorithm and K-means clustering RBF algorithm are used. Results show that the classification of the different wine samples is possible using the response signals of the E-nose. For the three neural networks BP, improved RBF and K-means RBF, the correct classification rates are 100%, 83.3%, 83.3% to original data, and they are 95.83%, 83.3%, 83.3% after process with PCA. From the test of two alcohols, the correct classification rates can reach 87.5%. The overall results show that the two neural networks can be employed for classification of the different wine samples. The classification method of ANN & PCA is proved to be a rapid and exact identification measure for pattern identification.